The binomial inverse theorem (Woodbury matrix identity)
ProvedFamousTheorems.woodbury_identityThe binomial inverse theorem (Woodbury matrix identity). Let () and () be invertible matrices over a commutative ring, an and an matrix, and suppose is invertible. Then
A rank- update of can therefore be inverted by inverting only an matrix. The case is the Sherman–Morrison formula. It is used throughout numerical linear algebra, Kalman filtering, Gaussian process regression and statistics.
Formalization note. Mathlib's Matrix.add_mul_mul_inv_eq_sub, with invertibility expressed by IsUnit and ⁻¹ the matrix inverse Matrix.inv.
Preamble
import Mathlib
Formal statement
namespace FamousTheorems
theorem woodbury_identity {m n α : Type*} [Fintype n] [DecidableEq n] [Fintype m] [DecidableEq m] [CommRing α] (A : Matrix n n α)
(U : Matrix n m α) (C : Matrix m m α) (V : Matrix m n α) (hA : IsUnit A) (hC : IsUnit C)
(hCVAU : IsUnit (C⁻¹ + V * A⁻¹ * U)) :
(A + U * C * V)⁻¹ = A⁻¹ - A⁻¹ * U * (C⁻¹ + V * A⁻¹ * U)⁻¹ * V * A⁻¹ := by sorry
end FamousTheoremsSource
Listed in Mathlib's curated theorem manifests (docs/1000.yaml); formalized in Mathlib as `Matrix.add_mul_mul_inv_eq_sub`. Proof here reduces to that Mathlib result.
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.